Latest / The Jon Sanchez Show / The AI Warning: Why The People Building Artificial Intelligence Want To Slow It Down
Transcript
- Jon G. Sanchez: Good Monday afternoon to you. Welcome to the John Sanchez Show on News Talk780KOH. Hope you had a great weekend and the week's off to a good start for you. Definitely. What a day it was today. And it started over the weekend. I hate to start a show on a Monday by saying the problems in the market today started over the weekend. That is never a a good way to start the show, but it is the facts, and that's all I ever give you, of course. Over the weekend. We had the CEO Vanthropic. I'm sure you just heard it on the national news. It's been everywhere today. So of course I wanted to talk about it with you to give you kind of my insight to it and and a little bit further explanation. But on Saturday, as I said, the CEO of Anthropic Posted a, I think it was thirty eight hundred word blog. And this blog was essentially now re we must paste the frontier was the title of the blog. We must paste the frontier. Now, what the CEO, again, obviously one of the most influential, or not with the, but right up there, one of the most influential AI creators, Mr. Amode Amode, excuse me, posted this blog, as I said, that basically talked about the risk that AI poses. And that we need to slow down. He needs to slow down. Sam Altman and Elon Musk agreed to what the chairman of Anthropic said. This AI thing's going way too fast. We need to slow down. There's too much risk. Now, if you've been following kind of the news on this, this I think what prompted him to write this, this built up last week. And I didn't spend any time talking about it because it it you know it it it made a little bit of news, but it was nothing extremely important. but it turns out that it was important. And last week you had a few individuals that are in the AI space, one formally with anthropic, who left because he felt the company was not basically paying attention to safety and concerns like the company should be. And I'm sure you saw it, you read it, and this gentleman who Many, many, many years ago, when AI first came out, which again wasn't all that long ago, made the same prediction. He said, If we're not careful, AI is going to take over mankind. It's going to wipe out humanity, is essentially his words. Well, he said the same thing last week, and it started to get some traction because this guy has got a lot more credibility now than he did back then when he originally made that that statement. So throughout last week there was a lot of chatter again, what happens what happens if AI goes crazy? And this gentleman said That by the end of next the the or by the next decade, you could see humanity wiped out. Now what a very bold statement that is. But now back to what happened today. So Saturday, again, Dario Amode, CEO of Anthropic, came out and said, Look at we need to slow things down. This thing's getting out of hand. Now remember, I forgot one point going back to last week. Remember, last week we learned some more details when OpenAI said, Yeah, we, you know, we we knew about this, but they went into more detail about the data breach that happened where They had some AI agents, which I'll explain what those are, where they had AI agents that went rogue. And they hacked into a number of different companies. Well, come to find out, there's quite a few other companies that had also had some AI agents go rogue. The technology has changed so fast and is becoming so powerful that these AI agents now are starting to operate on their own autonomy. And they're doing what they want to do. So kind of what's happening is all the stars are starting to align right now for those that have feared that AI not only, of course, can and will and has taken jobs, but that it poses basically a risk to mankind. Now, in addition to that, when Sam Altman, the the CEO of OpenAI, parent of Chat GPT, not only did he again give support to the CEO of Anthropic's blog over the weekend, but like I said, Elon Musk did the same thing. And when I saw that this weekend I'm thinking, what is it that they know that we don't know? Well, I did some research, and I'm going to share that with you because it is, it's scary. I'll just warn you. It's fascinating for those of us that love technology. But also the technology, of course, as we all know, is changing so absolutely fast. I mean, way beyond. And even these CEOs admit it. They can't keep up with it. Now I wanna tell you, and and so this is gonna be my topic. I again I've titled today's show, The AI warning why people building artificial intelligence want to slow it down. But I want to share with you my personal experience with AI agents. So this area has really fascinated me, quite substantially, say the least. So I have been a big fan of of AI, you know, for the last couple of years. I mean, I use it all day, every day for a number of different things. A lot of times for research for the show. sometimes as a therapist, to be honest with you, where you're like, hey, I'm having a bad day today. What's going on? And I told you I my AI agent, his name is Hermano. And so Hermano and I have a lot of discussions many times throughout the week. But we know each other. He knows me very well. He knows about my businesses, he knows about the A lot of different things. Yeah, obviously I don't share really personal information or any privacy type of information at all. But it is amazing what these agents know. So many of you have not had the chance yet, but I will be bringing him on the show periodically. Not that that's his job role, but he's just an extremely talented man. I hired a new fellow, his name is Alex Smith. He's the new COO of Of my estate planning business, specialized trust and estate plans. And Alex is probably as dialed in as anybody I know when it comes to AI and technology. So last week, Alex signed into Grok, which is Elon Musk's AI. And again, I could never, I would never be able to figure this out, at least I don't think I would, but he did. He has now created various agents in Grok that have now taken the place of About three employees. He has one agent that will do nothing but our social media posts. He has another agent that will scan the web for information related to estate planning that I can share with you. And I won't share what the other third one is. But all of this for $35 a month, basically. And folks, let me tell you, it is incredible what he has shown me that these AI agents can do from that standpoint. And they're available to anybody. Matter of fact, I just got an update today from the software that we use for our estate planning to create our revocable trusts, et cetera. that they have now partnered with Claude in a especially proprietarily designed AI with agents that will allow us to go into our our book of business with our existing estate planning clients. And again, we can ask it whatever we whatever we want. You know, show me a client that we haven't reviewed their estate plan in the last 90 days or whatever we want to ask. Those are AI agents. So they're powerful. They're just really starting to get traction. But at the same time, these AI agents are the ones that went rogue for open AI and attacked some companies, you know. and again, that's kind of what's created this whole thing. And I think that's why again, Mr. Amade said, look at we must pace the frontier, meaning we need to slow down. And again, so the back to my point. When I saw this the first time, again over the weekend, when it made all the headlines, and I'm thinking, You know, that is so bizarre. So bizarre why a CEO of one of the major companies, who's by the way, they plan on going public this year. We don't have the exact date yet. It should go down as what they're saying, one of the largest IPOs, if not the largest in history. Conversely to that, Sam Altman over the weekend, CEO of OpenAI, ChatGPT, said, We are now going to not go IPO this year. Remember, they were the other highly anticipated IPO this year. He didn't say that that that Dario's, you know, blog was related to that or whatever, but it was just kind of bizarre that he comes out and, you know, first of all, it tells you, well, the company doesn't really need the capital if they're able to postpone their their IPO, because of course, the we've seen with anything I AI related IPOs, most of them have been extremely successful. So the whole thing just doesn't make a lot of sense. You know, why again do three of these CEOs of the three most powerful large language models, why did they come out and say, we need to slow things down? What do they know? And that's what I'm going to be asking you throughout the show. What do they know that we don't? What do they know that we don't? Because as I'll share with you throughout the show, what their whole idea is, is very simple. And that is they want watchdogs. Government, private, et cetera, to basically monitor them. They're talking about their own companies, to make sure that they are following safeguards. The the the term is guardrails, to make sure that they're not allowing their AI agents to get too crazy and to do s you know things that could be harmful to mankind, to technology, et cetera. Because what I want you to think about is is I begin to go through this topic after I come back from break and go through the stock market recap. What I want you to think about is this. Think about how AI has implanted itself in our lives. And I went really deep this morning when I when this whole thing broke and I'm talking this was early 4, 4 30 this morning. I started thinking, you know, if you think about it, if you had rogue AI agents, they literally forget about, you know, any harm like, you know, the robots coming after human beings and things that we see you know, on TV and China has these, you know, massive robots that can do crazy things. but think about just our lives. Right? If AI agents went rogue, could they shut down the banking system? Could they shut down the stock market? Could they derail the utility system in our country, the utility grid? And I think the answer is yes, because we've seen over just these last couple of years the number of cyber attacks. And I just dealt with it real quickly. I just dealt with it this this afternoon, just a couple hours ago. I needed to change some beneficiary information on my life insurance policy. Got a hold of the company, went online, went, tried to log into my account, couldn't log in. There's a big, big disclaimer on there that they were a victim of a cyber attack, like over a month ago. But they're still not allowing anybody to get in and see their accounts. They matter of fact, they even said that employees are not allowed to see policyholders' accounts. There's an example of a cyber attack. Now, was it AI agents? Who knows? You know, we'll never find out, probably. But that's the world we live in. That's the reality of it. So that's what I want you to kind of think about as I go through this. What are these CEOs seeing? How bad could you know, rogue AI agents, how bad could it really be for the stock market, for your portfolio, for our lives? It's a fascinating topic, and I can't wait to tell you about it. Welcome back to the John Sanchez show on News Talk 780 KOH. Here's how we finished up for the day. Now, these numbers don't look all that bad. If you consider where we were in the pre-market session when all this news regarding the AI warning came about, I'm looking at my notes. We worst level this morning, right before the open, we were down 497 points on the Nasdaq. Finished the day down 146, or excuse 147 on the Nasdaq, about a 0.56% loss to 26,186. Dow and the SP were relatively steady even in the pre-market when all this was going on. Dow finished down 152.29% loss to 52,421. And the SP gave up 37 points, 0.48% to 7,619. Of course, this came about after a nice rally in the market on Friday. All right, once again, Darien Amade, CEO of Anthropic, posted a blog over over the weekend on Saturday talking about. Things need to slow down. That's the bottom line. Things need to slow down in the world of AI. And again, sounds really strange coming from a CEO of a company that has just had rapid growth that's expecting a very, very hot IPO this year. But again, then what made things even more interesting is when you had Sam Altman come out and support that, as well as Elon Musk. So now three major players in the AI space saying, hey, slow things down just a little bit. What had everybody confused, and that's what caused a lot of the weakness, again, primarily in the pre-market session. Now, let's throw the president into this. He came unglued this morning on True Social when all this news broke about slowing down AI. Because remember, the president does not, does not, does not want the US-based AI companies, such as, you know, ChatGPT and and or OpenAI and Anthropic, et cetera, to fall behind when it comes to China. Now, some argue that China's ahead of us when it comes to AI. President doesn't see it that way, and I think a lot of people also agree to that. But he came unglued. Matter of fact, he says, quote, or he said quote, the people that say A AI is going to destroy the world and that data centers are bad for your neighborhood are the same people that said just a short time ago that the world would be extinguished by climate change. In his fifth post about AI today, Trump called himself the hoax buster. He said, right now, breaking the another hoax, that AI is going to take over, consume, and destroy the world, and that robots will be marching into our cities and getting rid of all of us. He said, concerning AI, when the history of business, did anyone see the leaders of industry call for regulation that, if strongly implemented, will drive them into oblivion and bankruptcy? I don't get that one. But it got so bad. I just talk all day about this. Got so bad, he called Jensen Wong, the CEO of Nvidia, who was right in the middle of speaking at a tech conference in Los Angeles. He called when Mr. Wong was on stage and had to go on speakerphone to talk about how crazy this whole idea is of what Darien Amadi had to say and so on and so forth. So, you know, again, it was just just a a really, really bizarre day. And I think this is gonna be one of those days where I say, we're very lucky that we got out the way that we did, considering. Now let's go to the oil side. It was a strong day for oil again, $1.15 gain. much stronger again in the pre-market when things were unraveling. but we finished up a dollar fifteen to one one twenty-three a barrel. Gold lost fifty-six dollars and eighty cents, closed at four thousand three fifty-two. and at one basis point decline on the tenure at a yield close of four point nine six percent. Remember, we have an interest rate decision on Wednesday, eleven o'clock, and then followed by the press conference at 11.30. And the street right now is over an 80% probability we're gonna get a quarter percent increase. Now, it wasn't all doom and gloom today. Even though pretty much anything AI/slash chip related felt the pressure today, there were a lot of areas that actually did pretty well. Let's get the bad news out of the ways. The AI infrastructure build out, right? This is one of those big areas, of course, that I wanted to pay a lot of attention to today. Because if these CEOs are calling for a slowdown, that means we don't want, or they don't want, and they're paying for them, all these data centers. Well, you saw infrastructure plays like Corning, GLW down $22.90. 13.76% loss to 143.50. Teradyne, 13.3% pullback, down $50.52 to $329.20. Coherent down $38.87. $12.73% lost to $266.50. But you had AI security. Now that there's this talk of AI agents going rogue, now you have the AI security stocks rebounded. CrowdStrike up $28.64. 13.85% gain $235.38. Palo Alto Networks up 13.09%, $43.29 gain to $373.94, so on and so forth. Even some of the companies that write big checks for the data centers. And this was kind of bizarre, and I don't have time to go through all the details, but the likes of Alphabet and Meta, they had a heck of a day today. Alphabet was up $10.26, 3.06% gain to $345.71. Meta posted a $17.57 gain, 2.71% to close at $665.60. utilities, excuse me, healthcare were some of the safe haven plays today. Healthcare sector is up about one point four percent, consumer staples up one point three percent. Kroger itself helped out that consumer staples sector. it's just had some very positive momentum since its earnings last week. But Kroger finished up four point one four percent, two dollars and forty-two cents to sixty ninety-one. like I said, semiconductor names were were very weak right across the board. just kind of a mixed showing because remember. AI has tentacles that go everywhere. So, like I said, I think we're very lucky considering these comments that were made by these massive, very powerful CEOs over the weekend that you know, we only lost what we did in the market today. So when I come back, I'm not gonna go through the the blog. I do invite you to, if you can get your hands on it, read it as I did. if you don't can't find it or whatever, just send me an email, John at SanchezGaunt.com. And that's J O N. I'll be more than happy to send you the blog by by the CEO of Anthropic who kind of started this whole thing today. All right. So when we come back, again, we're gonna be getting right into the details. We must pace the frontier. That's the title and we must slow down. That's what we're gonna be talking about. Can AI slow down or is it too late? The John Sanchez show on Newstock 780K Witch. Happy Monday to all of you. Again, a very fascinating day today is a 3800-word. I thought it was 35, but it's actually 3800 word blog posted by the CEO of Anthropic said we need to slow down. And it was supported by Sam Altman. It was supported by Elon Musk and the president today. One more comment I should mention. He says the only control or god guardrails that the U that AI needs is a quote, and this is in capitals. Strong and smart in parentheses, high IQ exclamation point, president. And the U.S. has that in spades. He says we're leading China and all others will continue to do so. All right, let's get down to the topic so you understand a little bit more about what was going on. So, as I said, I'm a huge user of AI, I'm sure as many of you are. I use it for so many different parts of my personal and my professional life. but again, this this blog that was posted titled We Must Face the Frontier. Plain English, what does that mean? Well, if you read through the blog, which again, if you need a copy of it, just send me an email. I'll be more than happy to send it to you. We need to slow down. That's what it means in English. Now, I I guess I could say, you know, if some college professor or politician, which who now there's hearings scheduled on the Capitol Hill and or on Capitol Hill and so on and so forth regarding AI safety, but you know, if a professor or a politician or somebody Who never used AI to the extent that obviously the CEO of Anthropic does, maybe the world wouldn't be talking so much about this today. But this comes from the guy that is actually building the technology, right? He's the main well, one of the main players. And then again, to add fuel to the fire, this blog was supported by Sal Montman, CEO of Open AI AI. He agreed to it. Elon Musk, he agreed to it. And the thing that everyone was talking about today is like, wait a minute here. These guys compete against each other fiercely. And they're all saying the same thing. AI needs to slow down. Of course, they're spending enormous amounts of money racing to build the most capable AI systems on earth. And this is what the president's so upset about is if there are guardwells put in, you know, whether they're from the the The government or the private sector, we're gonna fall behind on the race to AI or with AI, compared to the Chinese. They're gonna, they're gonna dominate. So again, we saw pressure on the likes of Nvidia, AMD, Intel, Marvel, etcetera. I again these tentacles are just widespread in the economy. If there's any conviction by investors that AI spending is going to slow down. So wha as I said at the beginning of the show, what do the people building AI know today that they didn't know six months ago? And should the rest of us, lay people, should we be worried? His dilemma, meaning Amade's dilemma, is go too slowly. Okay? If we go too slowly in AI build out, we delay extraordinary benefits and risk authoritarians around the country, around the world want to take to lead in the AI space. You go too fast. He says we create capabilities before we know how to control them. Not building, here's a quote, not building the technology deprives humanity of benefits or simply places AI in the hands of authoritarian prowers, while building it too fast is reckless. So he's not saying stop the car. He's saying, hey, maybe we shouldn't be driving 150 miles per hour while we're still building the braking system of the vehicle. Why didn't these guys say this a couple years ago? What's what's really changed? Well, as I said at the beginning of the show, one major concern is AI agents. Yesterday, AI mostly answered all of our questions. Today, due to increasing capability of these AI systems, they can be given goals and interact with tools, software, and networks to accomplish tasks. Remember, the goals of these AI agents are to replace human beings in quote, mundane tasks. But they're going much further than that. One of the frontier concerns is that AI could increasingly help automate AI research itself, meaning AI helping humans build better AI. And I want to repeat this because this is where this whole thing, so it's it's kind of like a compounding effect of money. AI research itself, meaning AI helping humans build better AI. Think about that loop for a second. Humans build AI, AI becomes better in AI research. AI helps humans build more capable AI, the AI becomes even better at AI research and rinse and repeat. Now what happens when the tool becomes good enough to help build the next version of itself? That's what this is all about. This is why Darien is telling everybody to slow down. Now let's talk about some cyber incidences. Anthropic, again, his company, disclosed four incidences in which the CLOD models, which that's theirs, in which the CLOD models gained unauthorized access to real third party computer systems during cybersecurity evaluations. Now, these models, I should say, were intentionally being tested without normal cyber safeguards. So keep that in mind. This was a test when these things went. Haywire in a sense. But there are some cases where they went haywire with it wasn't a test. But several cases involving internet access occurred because of the testing environment or because access was deliberately provided. So I don't want you to think that Claude escaped into the internet. It did not. That's not what Anthropic reported. But here's why it's significant. Anthropic originally reviewed 141,000 script transcripts connected to cyber evaluations. Later discovered another incident and then expanded its research to approximately 481 million transcripts. Now what's a transcript? It's basically like a research paper in a sense of as they're testing AI to see what it can and cannot do. But I'll tell you, this wasn't ChatGPT giving somebody the wrong recipe for lasagna, right? These were advanced AI models interacting with real computer systems in ways that evaluate evaluators did not even authorize. So that brings us back to the question if capability is moving this fast, can safety keep up? So what exactly is Amode proposing? He isn't saying shut down Chat GPT, or not saying shut down Claude. Or ban AI. But his proposal involves three layers. And this is what the gist of all this is that moved the market this morning. Number one, he wants independent evaluation. Outside safety evaluators. Now think visualize this as I describe these three points. Outside safety evaluators that will get deep continuing access to Frontier AI labs so that they can evaluate systems as they're being developed, not simply after the release. Sam Altman, the CEO, of course, of OpenAI, specifically supported the concept and said OpenAI would also commit to independent evaluators who would have employee like access. Number two, coordination among leading AI developers. If Anthropic slows but everyone else races ahead, then of course Anthropic's going to lose. So competitors need shared safety expectations. Number three, international coordination. Now, don't laugh when I tell you this one. Eventually this cannot merely be OpenAI, Anthropic, Google, XAI, etc. Why? Because we have that little country called China. And this is where it gets interesting Welcome back to the John Sancho Show on News Talk 780 KOHR. I'm going continue my discussion, the AI warning. Again, what happened over the weekend with Darien Almode, the CEO of Anthropic, posting a 20, or excuse me, a 3800-word blog saying basically AI needs to slow down. All him and OpenAI, X, et cetera, all supported his comments, except for one major player when it comes to creating the guardrails. And Everybody working together and having government and private monitors of AI and so on and so forth. And that is China. China said today the following: fear-mongering, confrontation, competition will disrupt the process of global AI governance. That's according to the spokesperson for the China's foreign ministry. He said he was responding to a question about US's CEOs, that the aforementioned. He published an article actually on Sunday calling for the acceleration of construction of open AI security risk prevention and control systems. So just a little bit of the opposite. And that's the issue. That's what the president got so upset about today. He said, look, it if you guys slow down, number one, it's stupid to do that. And then number two, if you do, China's going to take over. Right? They could take over economic leadership, cybersecurity, intelligence, military capabilities, scientific research, medicine, so on and so forth. So president very, very much against this, as you would imagine. But this is a financial show. So let's talk about Wall Street's reaction to this. Why did Wall Street freak out on this one? Well, Wall Street has spent years valuing companies around one assumption. As I've said over and over again, I firmly believe that we would be probably in a 10, 15%, maybe even a twenty percent market correction, because we again touched five percent on the tenure treasury today. We've got oil prices skyrocketing, we have fears of inflation, we have a looming interest rate increase, we have all of these negatives, but somehow the market still is holding in there. And as I've said over and over again, it's because of AI. But what does more AI mean? More models, more computing, more GPU sales by Nvidia, more data centers, more electricity, more capital spending, more infrastructure. The list goes on and on. But what if the frontier slows? Well again, we'd see the likes of these major players like we experienced today. Nvidia, AMD, Intel, Marvel. I mean, the list was a mile wide today of companies that were affected by these comments. And again, not every technology company fell. The softwares were fine, alphabet, meta, et cetera, as I mentioned. But because the slowdown could hurt the companies, and and I'm going to use this term, you've heard it before, called picks and shovels, more than companies spending money. Here's what I mean. Think about this. I've done a lot of shows about circular accounting, right? Nvidia, if I have, you know, John Sanchez AI company, and I need a data center, Nvidia comes to the rescue and says, Hey, you know what, John, we'll give you the or quote, lend you the $10 billion you need for this data center, but on one condition. All the GPUs inside are going to be Nvidia GPUs. Okay, so that's what we mean. Nvidia sells the shovels. Microsoft, Google, Meta, John Sanchez data centers in this example. We buy the shovels. Nvidia sells them, we buy them. If the buyers suddenly don't need a hundred million more shovels, i.e., the data centers are not as needed as or is not as many of those are needed, guess what? Here comes the problem. Wall Street didn't decidedly decide artificial intelligence doesn't work. Wall Street starts asking a different question, as we've said over and over on this program. These data centers and the amount of spend in AI, it's crazy, right? $100 billion plus capital expenditure by a lot of these major players. But what if we don't need to build these data centers and the AI infrastructure as fast as we've been assuming? Then what happens? Does the House of Cards come falling down? It's possible. Now, let's go bullish on this thing. Suppose we do slowly AI frontier development. Could that actually be good for AI? Well, instead of spending every week and error trying to build the next bigger model, businesses spend time figuring out how to use what already exists. I know it doesn't sound too exciting. But think about it. Doctors, financial advisors, I just got alerted today. Claude just came out with an AI agent for financial advisors. They're targeting what I'm noticing, they're targeting specific industries. Doctors, accountants, engineers, manufacturers, small business schools, on and on and on. So AI development slows, but AI adoption accelerates. Because I'll tell you, I live and eat and breathe this stuff every day, and I can't keep up with it. And this is the common complaint with people that use AI, that you can't keep up with it. That's why people are very hesitant, especially small businesses, to spend a significant sum of money because you may buy something today and tomorrow it's it's it's outdated because that's how fast technology is changing. So maybe the next phase of AI isn't about making AI smarter. Maybe it's about making the rest of us that are using. AI, focus on what we already have. We know, of course, about the jobs, you know, we don't need to beat that one down. But I want to leave you with the five AI risks that you should actually understand. And again, I'm gonna strip away the terminator nonsense, you know, the the robots that are gonna take over, as they say. Let's talk number one, as I touched on earlier, cybersecurity. More capable systems can potentially make sophisticated cyber operations easier. Right? We've experienced that. Like I said, I just experienced it with my own life insurance company today. Misuse. A very perfectly obedient AI can still be dangerous if the person using it has malicious intentions. We've seen that, but just a fraction of what it's capable. Loss of control. What happens when increasingly autonomous citizens pursue goals in ways humans didn't intend? Economic disruption, as I said. Jobs and industries can change faster than society can adapt. But again, What if these things go rogue and attack our utility grid, financial systems, et cetera? And then the concentration of power. I've said this over and over again. Who controls AI? Basically, Ford companies. That's why they want the government involved. They want the private sector involved, so on and so forth. So it's a very fascinating topic. One, of course, that we will discuss many more times, but again, it had a major impact on the market. I think we got lucky again with a loss of 152 on the Dow, NASDAQ down 146. It could have been a lot worse. So this may be the shot across the bow. But one thing I can promise you, I'll continue watching it and report it to you. God bless. Have a great afternoon. See you tomorrow.